Download Evolutionary Approaches to Epistemic Justification

dialectica
dialectica Vol. 65, N° 4 (2011), pp. 517–535
DOI: 10.1111/j.1746-8361.2011.01283.x
Evolutionary Approaches to Epistemic Justification
dltc_1283
517..536
Helen De Cruz†, Maarten Boudry‡, Johan De Smedt§ and Stefaan Blancke!
Abstract
What are the consequences of evolutionary theory for the epistemic standing of our beliefs?
Evolutionary considerations can be used to either justify or debunk a variety of beliefs. This paper
argues that evolutionary approaches to human cognition must at least allow for approximately
reliable cognitive capacities. Approaches that portray human cognition as so deeply biased
and deficient that no knowledge is possible are internally incoherent and self-defeating. As
evolutionary theory offers the current best hope for a naturalistic epistemology, evolutionary
approaches to epistemic justification seem to be committed to the view that our sensory systems
and belief-formation processes are at least approximately accurate. However, for that reason they
are vulnerable to the charge of circularity, and their success seems to be limited to commonsense
beliefs. This paper offers an extension of evolutionary arguments by considering the use of
external media in human cognitive processes: we suggest that the way humans supplement their
evolved cognitive capacities with external tools may provide an effective way to increase the
reliability of their beliefs and to counter evolved cognitive biases.
1. The evolved mind and epistemic justification
Daniel Dennett (1995) has famously compared evolutionary theory to a universal
acid – a corrosive substance that eats its way through anything it touches, transforming every field it is applied to. Darwin’s idea of natural selection has the
power to affect ideas far outside its original domain, including economics, culture,
language and epistemology. Since evolutionary theory presents our current best
hope to explain design and adaptation from a naturalistic point of view, it is
perhaps not surprising that a growing number of philosophers (e.g., Quine 1969,
Fales 1996, Stewart-Williams 2005, Boulter 2007) incorporate evolutionary arguments in their naturalistic theories of mental content.
The aim of this paper is to examine the implications of evolutionary
epistemology for the epistemic justification of beliefs. The term ‘evolutionary
epistemology’ will be used in a broad sense, namely to denote the position that
†
Centre for Logic and Analytic Philosophy, Katholieke Universiteit Leuven, K.
Mercierplein 2, 3000 Leuven, Belgium and Somerville College, University of Oxford, Woodstock Road, OX2 6HD Oxford, United Kingdom; Email: [email protected]
‡
Department of Philosophy and Moral Sciences, Ghent University, Blandijnberg 2,
9000 Ghent, Belgium; Email: [email protected]
§
Department of Philosophy and Moral Sciences, Ghent University, Blandijnberg 2,
9000 Ghent, Belgium and Uehiro Centre for Practical Ethics, University of Oxford, Littlegate
House, 16-17 St Ebbes Str, Oxford, United Kingdom; Email: [email protected]
!
Department of Philosophy and Moral Sciences, Ghent University, Blandijnberg 2,
9000 Ghent, Belgium; Email: [email protected]
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
Published by Blackwell Publishing Ltd., 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350
Main Street, Malden, MA 02148, USA
518
Helen De Cruz et al.
biological evolutionary forces, in particular natural selection, are important in
shaping cognition. As we shall see in section 2, evolutionary considerations are
being used to either justify or debunk a wide variety of beliefs, including commonsense beliefs, religious ideas, moral judgments and scientific hypotheses.
However, as general strategies both justification and debunking are problematic.
The former might be subject to circular reasoning, e.g., using induction to justify
induction. The latter is potentially self-undermining – if our cognitive faculties are
deeply unreliable, why should we buy into evolutionary theory, which is after all
a product of those same cognitive faculties? Section 3 reviews and extends strategies to counter the circularity charge leveled against evolutionary arguments. In
section 4, we propose that incorporating the extended mind thesis in evolutionary
arguments can provide a means to justify beliefs, especially those outside the
scope of common sense. The example of temperature will illustrate how using
external media allows humans to reach beyond their evolved cognitive biases.
2. Cartesian God or Cartesian demon: the double-edged sword of evolution
Does the fact that the human brain is a product of organic evolution give us a
reason to believe that mental states accurately reflect the states of the world, or
does it lead to a far-reaching form of skepticism? Evolutionary approaches to the
mind have given rise to two mutually incompatible positions. The first position,
supported by evolutionary arguments (EAs), contends that natural selection
will tend to pick out and propagate those types of beliefs and judgments that
correspond with the state of the world. The second stance relies on evolutionary
debunking arguments (EDAs). An EDA is constructed by negating at least one of
the crucial EA premises, in particular about the relative importance of natural
selection, and about its truth-tracking ability. Although EAs and EDAs share
several premises, they reach contradictory conclusions. To see how they differ, let
us examine the general structure of both arguments (Figure 1).1
Note that 4* and 6* are logically independent defeaters of EA, and as we shall
see, several EDAs only make use of one of these premises. An EDA on the basis
of 4* emphasizes that natural selection is primarily concerned with fitness,2 not
1
This formulation of EAs and EDAs is general, in order to capture the similarities and
differences in the premises of both arguments. For an alternative formulation of EA, see Boulter
(2007); for another rendition of EDA, see Kahane (2011).
2
Several concepts of fitness are used in biology and in philosophy of biology (see, e.g.,
Ramsey 2006). Realized fitness tracks the actual reproductive success (number of offspring) of
individuals within a population. Propensity fitness conceptualizes fitness as the propensity of an
individual organism to produce a number of offspring. There can be a discrepancy between an
organism’s propensity fitness and its realized fitness. For example, an animal can have beliefs
that increase its propensity fitness (e.g., be properly cautious of predators) and yet fail to realize
this high propensity fitness because it is struck by lightning before reaching maturity. For the
purposes of this paper, we will use ‘fitness’ to mean propensity fitness.
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
Evolutionary Approaches to Epistemic Justification
519
Figure 1. Generalized forms of EA and EDA.
with tracking truths. According to EAs, true beliefs are better guides to action than
false beliefs, whereas EDAs that make use of premise 4* (e.g., Plantinga 1993) see
no reason why true beliefs would be privileged. By contrast, an EDA based on 6*
argues that natural selection is just one among many evolutionary forces, and there
are lots of factors that may interfere with it, giving rise to non-adaptive traits
(see subsection 2.2). EDAs that make use of premise 6* (e.g., Stich 1990) typically
grant that natural selection can sometimes be truth-tracking, but are skeptical of its
prominence in shaping cognition. They argue that alternative evolutionary explanations, if anything, are more detrimental to truth-value than explanations that
invoke natural selection. Thus, an EDA is an argument where one or both of the
crucial premises of EAs are negated. In order to assess the plausibility of these
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
520
Helen De Cruz et al.
positions, we will now look in more detail at the case for EAs and EDAs,
examining their scope and potential problems.
2.1. Evolutionary arguments
The wide applicability of EAs is potentially problematic, as it can lead one to
replace the Cartesian God with natural selection. Even if we assume that there is
some link between the truth of a belief and its adaptive value (premise 4), that
beliefs are partly though indirectly under genetic control, and that natural selection
is the primary drive of evolution (premise 6), the scope of EAs should be fairly
restricted. EAs have been put forward to justify a wide range of beliefs and
belief-producing mechanisms, such as the ability to make inductions (e.g., Quine
1969, 126), and to draw inferences to the best explanation (Goldman 1990). The
most successful EAs offered to date are in the domain of commonsense beliefs.
‘Common sense’ refers to those types of beliefs that are obvious and self-evident
to the subject, such as the existence of other minds, the occurrence of past events,
and the reliability of perception. Stephen Boulter (2007) argues that these are the
kinds of beliefs that are most likely to have a bearing on an individual’s fitness.
Echoing Thomas Reid (1764), he insists that commonsense beliefs are crucial for
our ability to act adaptively in the world. Their adaptive value provides evidence
for their validity. Steve Stewart-Williams (2005) focuses on one particular commonsense belief, our robust metaphysical belief in an observer-independent world.
People are typically not swayed by skeptical arguments that cast doubt on this
belief, and even if they proclaim to be skeptical, in practice they still intuitively
take the existence of the world to be self-evident. He argues that this deep
metaphysical conviction has an innate basis, and that the best explanation for this
is that the observer-independent world actually exists. In this way, EAs provide an
answer to the radical skeptic or solipsist, who denies the existence of the external
world:
The fact that any normal mind automatically assumes an objective and mindindependent external world may count as proof that such a world does exist. We
evolved a mind/brain that creates a sense of an objective, mind-independent external
world because this tendency generally contributed to the persistence of the genetic
material that gave rise to the tendency. In what kind of world would this tendency be
biologically advantageous? It would be advantageous in a world that genuinely exists
beyond our fleeting sensory impressions. The fact that this tendency evolved indicates that it was useful, and the simplest explanation for its usefulness is that it is
accurate (Stewart-Williams 2005, 794).
Critics have often complained that EAs are inherently circular, because such
arguments rely on a theory that is the product of human rationality to justify the
rationality of our beliefs. Thus, a naturalized epistemology based on evolutionary
theory is an inherently unstable position (Plantinga 1993). We will examine two
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
Evolutionary Approaches to Epistemic Justification
521
possible responses against this criticism in section 3. A second potential problem
for EAs is that they posit that successful (fitness-enhancing) action requires accurate beliefs. As we shall see in the next subsection, EDAs can cast doubt on this
premise.
2.2. Evolutionary debunking arguments
As argued above, EDAs attempt to undermine EAs by relying on one of two
possible strategies (or both): they attempt to disconnect the link between the truth
of a belief and its efficacy (4*), or they cast doubt on the shaping role of natural
selection in belief formation (6*). We will examine both strategies in turn.
Evolutionary theories of mental content propose that the ‘proper’ function of
our cognitive processes is to promote survival and reproduction (e.g., Millikan
1984). In analogy to the proper function of organs, such as the heart’s function of
pumping blood, our cognitive apparatus guides us to perform those kinds of
actions that enhance fitness. According to EDAs based on 4*, this should lead us
to doubt the accuracy of our beliefs. Natural selection may, for instance, produce
risk-aversive and thus error-prone cognitive mechanisms, as in the detection of
predators (see below). Natural selection is concerned with fitness, i.e., a propensity
to produce a greater number of surviving offspring, not with truth (premise 4*).
Alvin Plantinga (1993, 218), for instance, writes:
If our cognitive faculties have originated as Dawkins thinks [i.e., through natural
selection], then their ultimate purpose or function [. . .] will be something like
survival (of individual, species, gene or genotype); but then it seems initially doubtful that among their functions – ultimate, proximate or otherwise – would be the
production of true beliefs.
Interestingly, Charles Darwin,3 in a letter to William Graham, also expressed
doubts about the justification of our beliefs based on premise 4*:
With me the horrid doubt arises whether the convictions of man’s mind, which
has been developed from the mind of the lower animals, are of any value or at all
trustworthy. Would anyone trust in the convictions of a monkey’s mind, if there are
any convictions in such a mind? (Darwin 1881)
But even if premise 4 of EA is secured, establishing a connection between fitness
and truth-value, EDA proponents may try to undermine premise 6. Stephen Stich
(1990, 56) points out several reasons for why the powers of natural selection are
limited (premise 6*), so that evolution does not always produce “good approximations to optimally well-designed systems”: the fitting mutations may fail to
3
In his Descent of man, and selection in relation to sex (1871), Darwin argued that
human mental faculties are a product of natural selection, so that his doubt must have arisen from
premise 4* rather than 6*.
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
522
Helen De Cruz et al.
arise at the right time, or pleiotropic effects4 and drift5 may result in cognitive
mechanisms that are not fitness-enhancing. Even if the right mutations do occur, it
is possible that natural selection gets stuck on a local peak in a fitness landscape6
without being able to reach a better optimum. In addition, trade-offs may occur
between utility on the one hand and expenses in time and resources on the other,
resulting in cognitive systems that are less than reliable. Additionally, cognitive
mechanisms may also have come about through sexual selection, i.e., through the
interaction of phenotypes not with the environment, but with the fancy of the
opposite sex (Miller 2000).
Taking into account all these scenarios in which either premise 4 or premise 6
(or both) of EA fails to obtain, it seems that evolution has become more of a latter
day Cartesian demon. Plantinga (1993) is perhaps best known for his radical claim
that a naturalistic point of view is self-defeating, since evolution through natural
selection is not concerned with tracking truths, but with increasing fitness (4*).
He argues that only a supernaturalist ontology – where beliefs derive their warrant
from the fact that cognitive capacities have been designed by God in such a way
that they successfully aim at the truth – can provide an externalist justification for
our beliefs. However, most EDAs are less wide in scope than Plantinga’s.
Take EDAs against religious beliefs. Empirical evidence from developmental
psychology and cognitive science indicates that many elements of religious beliefs
arise early in development, and are stably present across cultures, such as an
overattribution of agency (we all see faces in the clouds (Guthrie 1993)), the
belief that minds persist after death (Bering 2006), and the intuition that natural
objects are designed for particular purposes (Kelemen 2004). What does this mean
for the epistemic justification of religious beliefs? Researchers in the cognitive
4
Pleiotropy occurs when one gene influences more than one trait. This may be problematic for selection, because the selection for one trait might favor one specific version of the
gene, whereas selection for another trait might favor another version of the gene. As a result,
neither trait will be optimal.
5
As in the case of fitness, the concept of drift has many connotations. It includes
indiscriminate sampling, e.g., large-scale events like floods that do not discriminate between fitter
and less fit individuals, as both types are equally likely to drown. In this example, fitness
differences cannot explain why a particular part of the population dies in an inundation, whereas
the other survives; it is a matter of being at the wrong spot at the wrong time. Another case of drift
is the founder effect, where part of a population gets isolated, leading to the spread of an
originally rare and non-fitness related trait in that subpopulation (see Walsh et al. 2002, for an
overview).
6
Fitness landscapes provide a visualization of the relationship between specific genotypes and reproductive success; peaks represent states where genotypes achieve a high realized
fitness. Rugged fitness landscapes have multiple peaks separated by valleys. In that situation, it
is difficult for a particular genotype to move away from one peak to reach another, higher peak,
because this move (crossing a valley) leads to temporary fitness costs. Organisms with a particular cognitive architecture may be stuck on a local, lower peak, as the costs involved in neural
reorganization may prevent them from evolving a more efficient brain.
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
Evolutionary Approaches to Epistemic Justification
523
science of religion have argued that religious beliefs are byproducts of cognitive
functions, such as agency detection and theory of mind (the ability to infer mental
states). In his attack on religion, Richard Dawkins uses the cognitive science
of religion literature to explain away religion as an accidental byproduct of our
cognition:
Religion can be seen as a by-product of the misfiring of several of these modules, for
example the modules for forming theories of other minds, for forming coalitions, and
for discriminating in favour of in-group members and against strangers (Dawkins
2006, 179).
But does being a byproduct by itself undermine religious beliefs? Dawkins does
not tell, but if that were so, one could argue that science, which is also a byproduct
of the evolved structure of human cognition, is likewise undermined. In both cases,
the trait in question would be an accidental byproduct of selection, not a direct
product of it (6*). One can formulate a byproduct account of religion in a more
refined way so that religion, but not other byproducts, is debunked. Paul Bloom
(2009), for example, acknowledges that the evolutionary origins of religious
beliefs do not provide a straightforward refutation of such beliefs. After all,
cognitive scientists also explore why people believe that 5 + 5 = 10, and none of
them come to doubt the validity of this outcome as a result. Nevertheless, he thinks
that the cognitive science of religion can challenge the rationality of holding on
to religious beliefs. Given that religion is an ‘evolutionary accident’, and that
a plethora of mutually incompatible religious concepts exists across the world,
believers may not be justified in holding the beliefs they do. Thus, it is not the
evolutionary origin of religion itself, but the fact that evolved cognitive biases give
rise to many incompatible beliefs, that provides a debunking argument against
religious beliefs. Similarly, Jesse Bering (2011) argues that, while evolutionary
theory does not disprove the existence of God, it nevertheless makes it improbable.
In principle, a theist could argue that God instilled religious belief in humans
indirectly, through natural selection, but “if scientific parsimony prevails [. . .]
such philosophical positioning [i.e., theistic evolution] becomes embarrassingly
like grasping at straws” (Bering 2011, 196). It is not the evolutionary origin of
religious belief in itself, but its conjunction with the principle of parsimony that is
used in a debunking argument. The case of religion indicates that evolutionary
origins of a belief do not always constitute sufficient grounds to dispel it.
Nonetheless, under specific circumstances, the evolutionary origins of a claim
may negatively affect its epistemic status. Thus, EDAs might work if they provide
fine-grained reasons for why the evolutionary origins of a particular belief might
affect its epistemic justification. We here provide four examples of specific EDAs.
i)
Cognitive processes may sometimes err on the side of caution. If the
costs or payoffs of false positives (detecting a signal in the environment
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
524
Helen De Cruz et al.
where there is none) and false negatives (failing to detect a signal that is
present in the environment) are asymmetric, natural selection will tend to
promote beliefs that yield the highest payoffs or incur the least costs
(Stephens 2001). An example of this is agency detection: humans and
other animals are prone to detect agency in the environment where none
is present (e.g., mistaking wind rustling in the foliage for an approaching
animal). This cognitive capacity generates an excess of false positives.
The evolutionary rationale for this is that a false positive is less costly
than a false negative, as the latter can result in a failure to detect a
dangerous predator, a prey, or a potential mate, and the former only
results in a small waste of time and energy (Guthrie 1993).
Animals are bounded in time and space which leads to trade-offs between
accuracy and efficiency. There is little point in carefully and elaborately
choosing the best escape route when faced with a hungry predator.
Some cognitive illusions may be adaptive (McKay and Dennett 2009).
The belief of devoted parents that their own children are more beautiful,
smarter and kinder than average (the Lake Wobegon effect) is clearly
unjustified as not every child can be above average, but it may contribute
to the time and energy parents invest in their children, thereby enhancing
their inclusive fitness. Wenger and Fowers (2008) found that the majority
of a sample of randomly selected biological parents of young children
holds unrealistically positive views about their children. The parents in
whom this illusion was most pronounced reported the highest degrees of
parental satisfaction.
Intuitive beliefs that have no bearing whatsoever on fitness are invisible
to natural selection, and are thus unreliable. After all, natural selection is
the only candidate for a truth-tracking evolutionary mechanism, but it is
of limited importance in shaping cognitive capacities. Next to this, we
can expect that evolutionary mechanisms such as drift (in the meaning of
stochastic processes) are even less truth tracking. Steven Pinker (2005)
speculates that this may account for the pervasiveness of cognitive biases
and illusions, such as the well-known conjunction fallacy.7
ii)
iii)
iv)
EDAs of types i–iii rely on premise 4*, as they emphasize the disconnection
between truth and fitness. By contrast, EDAs of type iv rely on premise 6*. Such
EDAs typically do allow for some connection between the fitness and truth-value
of a belief, but they are pessimistic about the extent to which natural selection
shapes our belief-formation processes. Let us focus on the former type first. EDAs
7
In the conjunction fallacy, the probability of two conditions in conjunction (A&B) is
regarded as (strictly) greater than that of a single one (A), which is incorrect according to
probability theory (Kahneman et al. 1982).
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
Evolutionary Approaches to Epistemic Justification
525
that primarily rely on premise 4* do not give us good reasons to believe in the
cogency of scientific reasoning, since they do not guarantee any link between
fitness and the truth-value of specific beliefs. This line of reasoning is selfdefeating. Evolutionary accounts according to which human cognitive capacities
are so deeply biased and defective that knowledge is ruled out are selfundermining. There would be no good reason to assume that scientific theories are
justified, or that philosophical reflection and argumentation (such as an EDA)
provides us with sound conceptual knowledge. Also, there would be no reason to
accept the soundness of psychological studies that indicate cognitive biases and
heuristics. The very fact that cognitive psychologists (e.g., Kahneman et al. 1982)
are able to recognize the frailties of human reasoning, and that subjects in psychological experiments perform better when their errors have been pointed out to
them, suggests that biases and heuristics are not so pervasive as to cloud our
reasoning completely, and that faulty reasoning is often corrigible with some
mental effort.
In the same vein, EDAs that rely on premise 6* cast doubt on any type of belief
that has no bearing on fitness. What kinds of beliefs would we be left with? The
relatively high reproductive success of people who are not scientifically literate
(like the Amish), or who actively oppose some forms of scientific knowledge (such
as fundamentalist Christians) suggests that scientific beliefs do not have much
impact on fitness (Kaufmann 2010). Likewise, there is little reason to expect that
our ability for philosophical reflection is subject to natural selection. But if scientific and philosophical beliefs belong to the type of beliefs that have no bearing on
fitness, and if EDAs indeed affect any such beliefs, then even fine-grained EDAs
that rely on premise 6* are potentially self-defeating. After all, these EDAs
themselves are based on scientific theories, notably evolutionary theory, and philosophical reflection. Thus, if we accept that everything that is not directly adaptive
or fitness-enhancing will be affected by an EDA, then EDAs of this type will
inevitably bloat out to a lot of other beliefs as well, undermining their own
coherence and leading one to doubt the cogency of scientific theories and philosophical argumentation. How do we avoid this self-defeat? One plausible solution
is that, although higher-order theories as proposed in scientific and philosophical
practice may be fitness-neutral, the cognitive skills they are based on need not be.
As David Papineau (2000) has suggested, selective pressures may have enhanced
human capacities for rational reasoning in the domains of folk psychology and
means-end reasoning, for example in our ability to discern causes. An EDA that
casts doubt on the adaptive value of these basic cognitive capacities (premise 6*)
is much less plausible. Indeed, as Evan Fales has argued, the probability that such
elaborate neural structures as are needed for these cognitive capacities would have
evolved without conferring any adaptive value is prima facie quite low, given their
high biological costs:
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
526
Helen De Cruz et al.
Homo sapiens has, more than any other species, specialized in intelligence as a
survival strategy [. . .] Our heavy investment in big brains and otherwise mediocre
bodies makes it all the more unlikely that resources would be wasted on elaborate
belief-forming and processing mechanisms that have no practical utility. (Fales 1996,
440)
If evolutionary approaches to the human mind are to be coherent, they should
allow at the very least for cognitive capacities that are capable of generating
truth-tracking theories, such as evolutionary theory. If we accept internal coherence as an important epistemic virtue, it seems that EAs are more promising than
EDAs in the formulation of a naturalistic theory of mental content, since the
former are not self-undermining. Therefore, the circularity charge, an important
challenge faced by EAs, should be addressed. In the following section, two types
of EA are presented that attempt to avoid this circularity problem.
3. Responses to the circularity charge
3.1. Dodging the bullet
One possible way out of the circularity problem is simply to grant the reliability of
our inductive capacities as given. This is an externalist position: we need not know
that beliefs are justified, it suffices that our cognitive processes are reliable to make
them justified. W.V.O. Quine is perhaps the best-known proponent of this naturalistic position:
I am not appealing to Darwinian biology to justify induction. This would be circular,
since biological knowledge depends on induction. Rather, I am granting the efficacy
of induction, and then observing that Darwinian biology, if true, helps explain why
induction is as efficacious as it is. (Quine 1975, 70)
In this strategy, the epistemologist simply refuses to address the second-order
question of whether our cognitive faculties are indeed reliable. This position is
committed to scientific naturalism, which argues that epistemological questions
should be approached through empirical science rather than through a priori
philosophy. A weakness of this strategy is the problem of cheap knowledge, which
is common to externalist positions in epistemology. Recall Plantinga’s (1993)
argument: if theism is true, we can expect that God has designed the human mind
in such a way that our cognitive capacities successfully represent true states of
affairs. Obviously, naturalists would object to this line of reasoning, but there
seems to be little to distinguish between the naturalistic explanation (by appeal
to evolutionary theory) and the theistic explanation (by appeal to Christian revelation). The naturalist can appeal to the primacy of natural science. Since
evolutionary theory is a scientific theory, it is – according to the naturalist – more
trustworthy than skeptical philosophical arguments. Thus, naturalized epistemologists can take evolutionary theory to be correct, and use this theory to explain why
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
Evolutionary Approaches to Epistemic Justification
527
our inductions are reliable. Similarly, the theist can appeal to the primacy of God
as first cause. However, this problem of cheap knowledge might be circumvented
if we consider naturalism not to be an a priori philosophical position, but
a historical, well-established, though provisional, result of scientific inquiry. In
contrast to supernatural accounts, naturalistic explanations have been consistently
successful. With the efficacy of induction taken for granted, and bearing in mind
the success of the naturalistic program in science, the naturalist can then argue that
evolution by natural selection indeed explains induction better than a theistic first
cause (Boudry et al. 2010).
3.2. Biting the bullet
Instead of simply refusing to address the problem of justifying our most basic
principles of reasoning, some authors have bitten the bullet and proposed ways out
of the circularity charge. F. John Clendinnen (1989) has argued that this part of the
project of evolutionary epistemology will be inevitably circular, but it need not
be viciously so. According to him, interdependence by itself is not sufficient to
establish vicious circularity. Consider the triplet of propositions ‘If P then Q’, ‘If
Q then R’, ‘If R then P’. Although this form of interdependence should make us
wary, the accusation of vicious circularity holds only when the circle is completely
closed, i.e., when we have no other reasons at all to accept any of the propositions
involved. In a deductive model of justification, it is often incorrectly assumed that
each step is either completely justified or not justified at all. However, it is possible
to have an interdependence of justifications (P, Q, R) in which we start out by
an initially very weak and provisional version of P, in which our confidence is
gradually raised as evidence accumulates (through Q and R). In particular,
Clendinnen (1989) believes that there are two reasons, independent from natural
facts, for accepting the principle of induction as minimally rational: the criterion
of non-arbitrariness and the principle of simplicity.
Clendinnen (1989) does not believe that we can justify the whole of the
scientific method a priori. Rather, he argues that, by starting from a minimally
rational principle of induction, we may accumulate evidence (including evolutionary theory) that itself lends support to the efficacy of induction: “[i]nduction itself,
once accepted as a minimal principle, may be used to interpret the available
evidence for or against the thesis that the world is the kind of place in which
induction is likely to succeed” (Clendinnen 1989, 468). Thus, by bootstrapping her
way out of the circularity, the evolutionary epistemologist is able to justify the
inferences on which her own belief in evolutionary theory hinges.
The Bayesian epistemologist Tomoji Shogenji (2000) proposes another way
out of the circularity problem. He argues that many forms of self-dependent
justification are in fact not circular – rather, they are unproblematic forms
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
528
Helen De Cruz et al.
of Bayesian confirmation. Standard Bayesian confirmation is a procedure that
strengthens the subjective belief in a hypothesis H by observation O relative to
background belief B if
1.
2.
The probability of H and B is strictly positive.
The probability of O given H and B is higher than the probability of O
given B alone.
In this general form of Bayesian confirmation, we compare Prob(O|H&B) and
Prob(O|B). If the former is higher than the latter, we can conclude that O confirms
H. Consider as an example empirical evidence for a scientific theory: empirical
evidence supports a theory if the evidence is more likely to occur under the
assumptions of the theory combined with our background beliefs than it would be
under our background beliefs alone. For example, the unusually slow precession
of Mercury’s orbit around the Sun (O) is in conflict with standard Newtonian
mechanics (N) and our background beliefs about the physical universe (B), but
this observation is predicted under the theory of general relativity (H). Thus,
this observation disconfirms N, Prob(O|N&B) < Prob(O|B), and confirms H as
Prob(O|H&B) > Prob(O|B).
Self-dependent justification differs from this general form in that H plays a
peculiar role, as it is both the hypothesis that is being tested and a part of the
background beliefs. If there is more to background beliefs than H, one can filter H
out of B to obtain B*. Thus, in such cases, the hypothesis is contained within the
set of background beliefs but it is not identical to the background beliefs. To apply
Bayesian confirmation to those cases, we need to replace B with B*&H. The
first of the two conditional probabilities then becomes Prob(O|H & B*&H). As
Prob(O|H&B*&H) = Prob(O|H&B*), the first conditional can be simplified as
Prob(O|H&B*). In this construal, H no longer plays the split role of hypothesis
and background belief, but rather, it plays the role of hypothesis twice in predicting
the probability of O. As H is no longer part of the background beliefs under this
construal, the proper second conditional becomes Prob(O|B*).
Shogenji (2000, 294) applies this model to evolutionary epistemology. Here
H stands for “perceptual process P is reliable”; O represents “S believes that P is
reliable”, and B* stands for “when it is used in empirical investigation, perceptual
process P generates a belief in the perceiver, to which she has an introspective
access to form a metabelief; her introspection, memory, etc., which do not depend
on her perception, are reliable”. In this case, O (S’s belief in the reliability of her
perceptual processes) does confirm H, because there is no reason from B* alone to
assume that S’s perceptual processes would be reliable; her perceptual processes
could generate any kind of belief, and it seems very unlikely that among those
beliefs would be her conviction that P is reliable. On the other hand, H confers a
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
Evolutionary Approaches to Epistemic Justification
529
high probability on those beliefs. In this way, the reliability of perceptual processes
can be tested by simple Bayesian confirmation.
Shogenji’s (2000) case for evolutionary epistemology could be strengthened.
The case as he presents it relies on a theory-dependent observation O. But current
evolutionary approaches to the human mind frequently appeal to observations that
are not dependent on evolutionary theory itself – of course, they are dependent
on other theories, but crucially, they do not depend on the specific evolutionary
hypotheses they set out to test. Such observations, which are neutral with respect
to the theory that is being tested, are denoted with O* (a term borrowed from Adam
(2004)). Over the past decades, evolutionary psychologists8 have used findings
from neuroscience and cognitive, comparative and developmental psychology to
test evolutionary hypotheses (see, e.g., Dunbar and Barrett 2007). Such O* types
of observation do not assume that evolutionary theory is true, and therefore can
confirm evolutionary hypotheses about cognition without circularity. Take, for
example, the observation in comparative psychology that humans and monkeys
are fast and accurate in their categorization of various kinds of stimuli. Rhesus
monkeys can reliably sort pictures into food and nonfood categories, even if the
pictures show items they are unfamiliar with (Fabre-Thorpe et al. 1998). Moreover, they can do this in a very brief period (pictures are flashed for a duration of
only 80 ms), and they are correct in approximately 90% of the trials. Next to this,
a wealth of experimental data from developmental psychology (e.g., Farroni et al.
2005) shows that human newborns have a visual preference for face-like stimuli.
Moreover, young infants can recognize their own mothers from other women with
similar clothes and hairstyle within a few hours after birth (Bushnell 2001). These
observations do not rely on the supposition that evolutionary theory is correct, but
they do strengthen evolutionary arguments for the reliability of our beliefs. The
categorization studies indicate that monkeys are capable of correctly categorizing
stimuli. Their proficiency given the hypothesis that natural selection has endowed
them with cognitive capacities that are at least truth preserving under some
conditions is more likely than their proficiency given everything else we know
about the physical world: the complexity of visual scenes, together with the lack
of previous experience with the test items leads to the expectation that the
monkeys would perform at chance level, Prob(O*|H&B*) > Prob(O*|B*). Pace
Darwin’s (1881) earlier-mentioned skepticism about monkeys’ minds, there are
indeed good reasons to trust a monkey’s convictions, at least when it comes to
discriminating food from nonfood. Likewise, the studies with face-recognition in
human newborns provide us with an observation that seems vastly improbable
given our background knowledge about their lack of experience (having spent their
8
Evolutionary psychologists examine cognitive capacities (especially those of humans)
as a product of functional design caused by natural and sexual selection.
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
530
Helen De Cruz et al.
time in the dark environment of the womb) and their poor visual acuity. By
contrast, the observation is more likely if we accept an evolutionary hypothesis
that proposes that humans are equipped with an evolved, unlearned capacity to
recognize faces. Such an ability would have been adaptive for primates living
in social groups in order to recognize each other, especially given that diurnal
primates like ourselves have less-developed olfaction compared to most other
mammals. In sum, circularity can be avoided by a careful rephrasing of EAs, and
by reliance on observations outside the domain of evolutionary theory. Nevertheless, it remains to be seen whether non-circular EAs could be constructed for other
cognitive processes, such as inductive inferences. After all, neuroscience and other
scientific disciplines do not deliver brute facts, but require interpreting, cognizing
minds. If these inferential processes are at stake, it is hard to see how one can
construct a non-circular EA.
4. Extended cognition and evolved cognitive biases
As we have seen, EAs are most successful for commonsense beliefs. What justification can we get for other types of beliefs, like scientific knowledge, mathematical results or philosophical argumentation, which may be outside the grasp
of natural selection – in other words, how can we counter premise 6* of EDAs?
Additionally, how can we counter premise 4* for cognitive capacities where we
may expect that natural selection has shaped cognitive capacities that do not track
truth? In subsection 2.2 we suggested that although scientific and philosophical
beliefs may be fitness-neutral, this does not imply that the cognitive skills on
which they are based would be fitness-neutral. A capacity like induction is useful
in a wide variety of settings, not just in the context of academic reflection.
Additionally, in section 3 we outlined several strategies that allow for selfdependent justification in EAs, and that avoid the circularity charge.
In what follows, we explore a further strategy to argue that humans can
have reliable beliefs despite cognitive biases. We will argue that the way humans
naturally supplement their evolved cognitive capacities with external tools may
provide an effective way to increase the reliability of their beliefs and to counter
evolved cognitive biases. Outsourcing cognitive tasks to the external environment enhances cognition in several ways. First, it improves conceptual stability.
Thinking sometimes involves complex manipulations of conceptual structures,
as in logical reasoning or carrying out large calculations. The stability of these
manipulations (e.g., substituting a constant by another in a logical proof) is
greatly enhanced by writing down each step (Clark 1996; De Cruz and De Smedt
in press). Second, extending the mind may provide a way around evolved cognitive biases. For example, a large experimental literature (see Loftus (2003) for
a review) indicates that people’s episodic memory (i.e., biographical memory of
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
Evolutionary Approaches to Epistemic Justification
531
personal experienced events) is highly constructive and liable to distortion. For
instance, people typically remember their worst train-missing experience when
simulating how painful and inconvenient a next train-missing experience will
be (Morewedge et al. 2005). This puzzling feature of episodic memory can be
explained by the hypothesis that its function is not one of disinterested representation of true events, but one of building simulations that guide our actions in
adaptive ways. Overestimating the discomfort of an unpleasant experience may
help us to avoid that situation in the future. Adaptive as this may be, it poses
severe limitations on the reliability of our long-term episodic memories, which
typically get distorted over time. Artifacts, such as books, journals, electronic
storage devices, measuring instruments, calendars or even simple tallies, allow us
to store information that is cognitively challenging to memorize, and to protect it
from memory distortion. In this way, an EDA that would call the reliability of
human memory (a commonsense belief) into question could be countered as
follows: human episodic recall may be biased, but humans can mitigate this by
using external memory systems.
The use of external media is not limited to contemporary societies, but seems
to be a pervasive element of human cognition at least since the late Pleistocene (ca.
120,000–12,000 years ago). From this period onward, archeologists find notched
pieces of ochre and bone, shell beads and representational art, demonstrating that
humans conveyed ideas externally in symbolic media (De Smedt and De Cruz
2011). Kim Sterelny (2003) incorporates theories of niche construction in human
cognitive evolution, arguing that, just as termites build their own environment,
humans construct their own cognitive niche using artifacts to suit epistemic purposes. Niche construction is the process in which organisms change their own
selective environments, thereby influencing their evolutionary history. Common
examples include termites, beavers and ants. Members of these species build
complex artifacts to ameliorate their environment in terms of temperature and
humidity, and to provide an optimal nursery setting for offspring (Laland and
Brown 2006). Similarly, humans improve their environment by building houses,
boats and other artifacts; they also improve their epistemic environment by
designing tools to suit their cognitive purposes. If the extended mind is indeed
a key feature of human cognitive evolution, it might be possible to justify noncommonsense beliefs through EAs, namely as those beliefs that reliably arise
through the judicious use of external media which can counter cognitive biases.
Of course, the use of external tools does not guarantee that the beliefs acquired
or stored in this way will always be reliable. It does, however, provide an evolutionary explanation for why humans can have knowledge that goes beyond their
commonsense intuitions.
Temperature provides an interesting case study of the biases of external
perception, and the use of external media in surmounting these. Peripheral
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
532
Helen De Cruz et al.
thermoreception is the system that reacts to surface skin temperature. As Kathleen Akins (1996) compellingly argues, this system does not provide us with a
reliable representation of the external world. Rather, it tends to produce representations that are especially conducive to an organism’s fitness. There are four
kinds of thermoreceptors in the skin: ‘warm spots’, ‘cold spots’ and two types of
pain receptors that fire under conditions of extreme heat or cold.9 The ratio
between warm spots and cold spots is not evenly distributed across the body, for
example, the lips are almost exclusively sensitive to warmth, and the scalp is
mainly sensitive to cold. Also, the sensitivity of these receptors is not a linear
function, and their response critically depends upon the starting temperature.
The change from tepid to warm water, for example, elicits less response than
that from warm to hot water. These curious properties of thermoreception can be
explained in evolutionary terms – what is important to warm-blooded creatures
like us is that our perception of temperature helps us to act adaptively in the
world, to avoid injury, overheating and hypothermia. Thus, it makes sense that
the human scalp is sensitive to cold, since excessive cooling of the brain is
potentially life-threatening. The fact that thermoreception is concerned with an
organism’s fitness, rather than with an accurate, disinterested rendition of temperature, makes it vulnerable to various biases. For example, if you place your
left hand in hot water and your right hand in cold water, they will register
different temperatures when put in the same lukewarm water. Seen from the
perspective of an organism that has evolved mechanisms to maintain a constant
body temperature, such illusions make sense.
Humans, being used to rely on external media, have come up with an elegant
solution to the inherent biases of thermoreception: find fixed points outside of
our own thermoreceptive experience and use these as a scale of reference for
external devices that register temperature. Already since the 1600s, scientists
who attempted to make reliable thermometers had a preference for observerindependent points, such as the melting point of butter or the boiling point of wine.
Hasok Chang (2004) provides a detailed account of the human quest for standardized temperature measurement and of the establishment of the (to some extent
idealized) fixed points of freezing and boiling water on the Celsius scale. To be
sure, the fact that we place epistemic trust in such external devices does require the
reliability of other intuitions and cognitive faculties, such as transitivity (if x is
warmer than y, and y is warmer than z, then x is warmer than z), and a trust in the
existence of general laws of nature. Nevertheless, relying on external measuring
devices provides a solution for at least one set of cognitive biases, namely those of
human thermoreception.
9
These data on thermoreception are derived from Akins (1996).
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
Evolutionary Approaches to Epistemic Justification
533
Once humans were able to outsource their measurement of temperature to the
outside world, they were not only able to measure temperature in a more objective
way, but also in a much more precise fashion than if they relied on their evolved
capacities alone. Interestingly, the practical advances in thermometry also contributed to the conceptual understanding of thermodynamics, such as the understanding that cold is not on an ontological par with heat, and the formulation of an
absolute zero point. The folk intuition that cold and heat are both equally real
phenomena may be derived from our tactile perception of temperature (the occurrence of cold spots and warm spots), an intuition that was countered by experiments that indicated that cold is just the absence of heat. The fact that current
thermodynamics flies in the face of our evolved capacities for thermoreception
provides at least prima facie evidence against a sweeping applicability of EDAs.
Humans are able to surmount the biases of their evolved thermoreception through
external tools. To put it differently: we have an intuition that cold really exists and
there is a good evolutionary rationale for this. In spite of this, we can come to
believe that cold is merely absence of heat – for this too, there is a good evolutionary explanation, namely our evolved ability to use external media in epistemic
contexts.
5. Conclusion: a defeasible evolutionary account
What are the consequences of evolutionary theory for the epistemic standing of
our beliefs? The development of EAs is faced with several challenges. Advocates
of EDAs point out that there is no connection between fitness and truth-value
(premise 4*) or that the force and scope of natural selection may be constrained for
several reasons (premise 6*). In addition, EAs are open to the charge of circularity.
We have explored venues for setting up EAs: the externalist solution, which uses
bootstrapping by way of benign as opposed to vicious circularity and the use of
observations that are independent from evolutionary hypotheses. Next to this, we
have proposed that incorporating external media in human cognition provides
us with a defeasible EA that indicates that knowledge outside of the domain
of common sense is at least a prima facie possibility. Truth-approximating or
instrumentally useful knowledge can be attained by reliance on external tools
that are independent of human cognitive biases. Of course, emphasizing the role
of external media in human cognition does not provide us with a universal,
all-encompassing epistemic justification for human knowledge. What it does
provide is a plausible reason, from an evolutionary perspective, of why humans
can have knowledge outside of domains where accurate beliefs matter for survival
and reproduction, like, for example, scientific knowledge. As we have seen,
EAs that do not appeal to the extended mind seem limited to the domain of
commonsense knowledge. The role of external media in human cognition can
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
534
Helen De Cruz et al.
provide an epistemic justification for some non-commonsense beliefs that humans
entertain.*
References
Adam, M. 2004, “Why Worry About Theory-Dependence? Circularity, Minimal Empiricality and
Reliability”, International Studies in the Philosophy of Science, 18, pp. 117–132.
Akins, K. 1996, “Of Sensory Systems and the ‘Aboutness’ of Mental States”, Journal of Philosophy,
93, pp. 337–372.
Bering, J.M. 2006, “The Folk Psychology of Souls”, Behavioral and Brain Sciences, 29, pp. 453–462.
Bering, J.M. 2011, The God Instinct. The Psychology of Souls, Destiny and the Meaning of Life,
London: Nicholas Brealy.
Bloom, P. 2009, “Religious Belief as an Evolutionary Accident”, in: J. Schloss and M.J. Murray, eds,
The Believing Primate. Scientific, Philosophical, and Theological Reflections on the Origin of
Religion, Oxford: Oxford University Press, pp. 118–127.
Boudry, M., Blancke, S. and Braeckman, J. 2010, “How Not to Attack Intelligent Design Creationism: Philosophical Misconceptions about Methodological Naturalism”, Foundations of
Science, 15, pp. 227–244.
Boulter, S.J. 2007, “The ‘Evolutionary Argument’ and the Metaphilosophy of Commonsense”,
Biology & Philosophy, 22, pp. 369–382.
Bushnell, I.W.R. 2001, “Mother’s Face Recognition in Newborn Infants: Learning and Memory”,
Infant and Child Development, 10, pp. 67–74.
Chang, H. 2004, Inventing Temperature: Measurement and Scientific Progress, Oxford: Oxford
University Press.
Clark, A. 1996, “Linguistic Anchors in the Sea of Thought?” Pragmatics & Cognition, 4, pp. 93–103.
Clendinnen, F.J. 1989, “Evolutionary Explanation and the Justification of Belief”, in: K. Hahlweg
and C.A. Hooker, eds, Issues in Evolutionary Epistemology, Albany, NY: State University of New
York Press, pp. 458–474.
Darwin, C. 1871, The Descent of Man, and Selection in Relation to Sex, London: John Murray.
Darwin, C. 1881, Letter 3230 – Charles Darwin to William Graham, 3 July 1881, retrieved from
http://www.darwinproject.ac.uk/entry-13230 on 20 May 2010.
Dawkins, R. 2006, The God Delusion, Boston: Houghton Mifflin.
De Cruz, H. and De Smedt, J. in press, “Mathematical Symbols as Epistemic Actions”, Synthese.
De Smedt, J. and De Cruz, H. 2011, “The role of material culture in human time representation.
Calendrical systems as extensions of mental time travel”, Adaptive Behavior, 19, pp. 63–76.
Dennett, D.C. 1995, Darwin’s Dangerous Idea. Evolution and the Meanings of Life, London: Allen
Lane.
Dunbar, R.I.M. and Barrett, L. (eds) 2007, Oxford Handbook of Evolutionary Psychology, Oxford:
Oxford University Press.
Fabre-Thorpe, M., Richard, G. and Thorpe, S.J. 1998, “Rapid Categorization of Natural Images
by Rhesus Monkeys”, NeuroReport, 9, pp. 303–308.
Fales, E. 1996, “Plantinga’s Case Against Naturalistic Epistemology”, Philosophy of Science, 63,
pp. 432–451.
Farroni, T., Johnson, M.H., Menon, E., Zulian, L., Faraguna, D. and Csibra, G. 2005,
“Newborns’ Preference for Face-Relevant Stimuli: Effects of Contrast Polarity”, Proceedings of
the National Academy of Sciences of the USA, 102, pp. 17245–17250.
* We express our gratitude to Anne Meylan, Igor Douven, and an anonymous reviewer
for their elucidating comments on and suggestions regarding an earlier version of this paper. This
research is supported by the Research Foundation Flanders and grants COM07/PWM/001 and
BOF08/24J/041 from Ghent University.
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.
Evolutionary Approaches to Epistemic Justification
535
Goldman, A. 1990, “Natural Selection, Justification, and Inference to the Best Explanation”, in:
N. Rescher (ed.), Evolution, Cognition and Realism. Studies in Evolutionary Epistemology,
Lanham: University Press of America, pp. 39–46.
Guthrie, S. 1993, Faces in the Clouds: A New Theory of Religion, New York: Oxford University Press.
Kahane, G. 2011, “Evolutionary Debunking Arguments”, Noûs, 45, pp. 103–125.
Kahneman, D., Slovic, P. and Tversky, A. 1982, Judgment under Uncertainty: Heuristics and
Biases, Cambridge: Cambridge University Press.
Kaufmann, E. 2010, Shall the Religious Inherit the Earth? Demography and Politics in the
Twenty-First Century, London: Profile Books.
Kelemen, D. 2004, “Are Children ‘Intuitive Theists’?”, Psychological Science, 15, pp. 295–301.
Laland, K.N. and Brown, G.R. 2006, “Niche Construction, Human Behavior, and the Adaptive-Lag
Hypothesis”, Evolutionary Anthropology, 15, pp. 95–104.
Loftus, E.F. 2003, “Make-Believe Memories”, American Psychologist, 58, pp. 867–873.
McKay, R.T. and Dennett, D.C. 2009, “The Evolution of Misbelieve”, Behavioral and Brain
Sciences, 32, pp. 493–510.
Miller, G. 2000, The Mating Mind. How Sexual Choice Shaped the Evolution of Human Nature,
London: William Heineman.
Millikan, R. 1984, Language, Thought, and Other Biological Categories, Cambridge, MA: MIT
Press.
Morewedge, C.K., Gilbert, D.T. and Wilson, T.D. 2005, “The Least Likely of Times: How
Remembering the Past Biases Forecasts of the Future”, Psychological Science, 16, pp. 626–630.
Papineau, D. 2000, “The Evolution of Knowledge”, in: P. Carruthers and A. Chamberlain, eds,
Evolution and the Human Mind. Modularity, Language and Meta-Cognition, Cambridge: Cambridge University Press, pp. 170–206.
Pinker, S. 2005, “So How Does the Mind Work?” Mind and Language, 20, pp. 1–24.
Plantinga, A. 1993, Warrant and Proper Function, Oxford: Oxford University Press.
Quine, W.V.O. 1969, Ontological Relativity and Other Essays, New York: Columbia University Press.
Quine, W.V.O. 1975, “The Nature of Natural Knowledge”, in: S. Guttenplan, ed., Mind and Language:
Wolfson College Lectures, Oxford: Clarendon Press, pp. 67–81.
Ramsey, G. 2006, “Block Fitness”, Studies in the History and Philosophy of Biology & Biomedical
Sciences, 37, pp. 484–498.
Reid, T. 1764, An Inquiry into the Human Mind, on the Principles of Common Sense, Edinburgh:
Millar, Kincaid & Bell.
Shogenji, T. 2000, “Self-Dependent Justification Without Circularity”, British Journal for the
Philosophy of Science, 51, pp. 287–298.
Stephens, C.L. 2001, “When is it Selectively Advantageous to Have True Beliefs? Sandwiching the
Better Safe than Sorry Argument”, Philosophical Studies, 105, pp. 161–189.
Sterelny, K. 2003, Thought in a Hostile World: The Evolution of Human Cognition, Oxford:
Blackwell.
Stewart-Williams, S. 2005, “Innate Ideas as a Naturalistic Source of Metaphysical Knowledge”,
Biology & Philosophy, 20, pp. 791–814.
Stich, S.P. 1990, The Fragmentation of Reason: Preface to a Pragmatic Theory of Cognitive Evaluation, Cambridge, MA: MIT Press.
Walsh, D.M., Lewens, T. and Ariew, A. 2002, “The Trials of Life: Natural Selection and Random
Drift”, Philosophy of Science, 69, pp. 452–473.
Wenger, A. and Fowers, B.J. 2008, “Positive Illusions in Parenting: Every Child is Above Average”,
Journal of Applied Social Psychology, 3, pp. 611–634.
© 2011 The Authors. dialectica © 2011 Editorial Board of dialectica.